Splicing quantification

MAJIQ has extensive functionality to quantify and assess changes in splicing. This functionality starts with the PsiCoverage files produced by majiq psi-coverage. For the quantifiable LSVs that this step identifies, MAJIQ infers posterior distributions of the Percent Spliced In (PSI) for raw and bootstrap replicates of LSV coverage as described in [Vaquero2016], [VaqueroAicherJewellGazzara2021].

MAJIQ provides three different quantification modes:

• majiq psi: summarize inclusion from a single group of experiments as a posterior distribution over PSI. Input experiments are either quantified independently or as replicates (depending on --min-experiments flag).

• majiq deltapsi: summarize differences in inclusion between two groups as a posterior distribution over a difference in PSI (dPSI). Input experiments are considered as replicates (one distribution on PSI per group, combined for one distribution on dPSI between groups).

• majiq heterogen: test differences in PSI between samples from groups. Input experiments are considered independent samples (one distribution on PSI per sample). Test for differences between groups using 2-sample statistical tests on posterior means and samples generated from posterior distributions.

These quantifiers generally produce an output TSV (to stdout, alternatively to path indicated by --output-tsv). One should also generally add the --splicegraph SG flag to annotate the quantifications with information about the LSVs being quantified. PsiCoverage is specified for majiq psi as positional arguments. For the other two quantifiers, specify group names with -n NAME1 NAME2 and PsiCoverage for each group with -grp{1,2} PSICOV [PSICOV ...].

Additional details on required/optional parameters can be found by adding --help to the subcommand of interest (e.g. majiq deltapsi --help).

Alternatively, these quantifiers wrap the PsiCoverage and quantifier APIs (in development), which can be used directly for more control over the analysis.